Heart failure & cardiomyopathies
A deep patient-similarity learning framework for the assessment of diastolic dysfunction in elderly patients
Authors' conclusion
Despite training with a younger cohort, a deep patient-similarity-based learning framework for assessing LVDD provides a robust prediction of all-cause death and incident HF for older patients.
Study context
- Population
- Older participants (66–90 years) of the Atherosclerosis Risk in Communities Study; ACCF/AHA Stage A/B and Stage C/DStage A/B 4054
- Participants
- 5,596 participants
- Exposure
- Deep neural network (DeepNN)-predicted high-risk vs low-risk phenogroup
- Outcome
- Incident HF and all-cause death
Main finding
As reported in the abstract
High-risk phenogroup vs Low-risk phenogroupAdjusted hazard ratio (95% CI)P value
HF, Stage A/B
6.52(4.20–10.13)
P < 0.0001Death, Stage A/B
2.21(1.68–2.91)
P < 0.0001HF, Stage C/D
6.51(4.06–10.44)
P < 0.0001Death, Stage C/D
1.03(1.00–1.06)
P < 0.0001- Increased incidence of heart failure (HF) and death in high-risk phenogroup, Stage A/B and Stage C/D (log-rank)
- P < 0.0001
Observational data: associations, not effects.
Further findings
C-statistic
| Outcome | DeepNN | ASE/EACVI | Net re-classification index (CI) | P value |
|---|---|---|---|---|
| Incremental value of DeepNN over 2016 ASE/EACVI guidelines | 0.76 | 0.70 | 0.5 (0.4–0.6) | P < 0.001 |
| Incremental value of DeepNN over 2016 ASE/EACVI guidelinesMaintained across stage groups | P < 0.001 |
Original publication
A deep patient-similarity learning framework for the assessment of diastolic dysfunction in elderly patients. Eur Heart J Cardiovasc Imaging. Published 5 February 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →